Making Better Estimates of Prospect Reserves

Peter R. Rose

Most geoscientists predict reserve-sizes of successful exploratory prospects poorly. They are overconfident as to possible ranges of reserves; strongly overoptimistic; and therefore frequently surprised, usually on the downside. So investors suffer, and exploration staffs lose credibility,

Prospect reserves (ultimate recoverable bbl/mcf) depends on three variables: Area of accumulation (ac); Average net feet of pay (ft); and Hydrocarbon-recovery factor (bbl or mcf/ac-ft). These variables are nearly always independent, and appear to be distributed lognormally. Their product, the prospect reserves distribution, commonly generated by Monte Carlo simulation, is also lognormal.

Prospectors can learn to make better forecasts of prospect reserves through routine use of six different but mutually supportive estimating techniques: (1) mapping both genesis and geometry of traps employing multiple working hypotheses; (2) avoiding common psychological biases, such as "anchoring"; (3) constructing distributions of Area, Pay, HC-recovery, and Reserves which are constrained by the expectation of lognormality; (4) applying "reality-checks" to estimates based on (a) known analogous or parent geological situations such as field-size distributions, (b) P_1%, P_50%, and P_99% reserves estimates that are consequent to the proposed log normal distribution and believable geologically; and (c) variance of reserves distribution that is consistent w th that class of exploratory well, and with prospect data quality and quantity; (5) combining independent estimates by multiple seasoned explorers, who have utilized identical prospect data-sets; (6) getting "feedback" through routine comparisons of prior predictions and actual outcomes.

AAPG Search and Discover Article #91019©1996 AAPG Convention and Exhibition 19-22 May 1996, San Diego, California